NILS-HENRIK M. VON DER FEHR EIRIK S. AMUNDSEN LARS BERGMAN

T H E N O R D I C M A R K E T: S I G N S O F S T R E S S? NILS-HENRIK M. VON DER FEHR EIRIK S. AMUNDSEN LARS BERGMAN March 17, 2005 The supply shock that hit the Nordic electricity market in 2002-2003 put the market to a severe test. A sharp reduction in inflow to hydro reservoirs during the normally wet months of late autumn pushed electricity prices to unprecedented levels. We take this event as the starting point for analysing some potential weaknesses of the Nordic market. We conclude that fears regarding supply security and adequacy are likely to be unfounded. Nevertheless, as inherited over-capacity is eroded, and new market-based environmental regulation takes effect, tighter market conditions are to be expected. It is then crucial that retail markets are fully developed so as to allow consumers to adequately protect themselves from occurrences of price spikes. INTRODUCTION The Nordic electricity market encompassing Denmark, Finland, Norway and Sweden 1 is well established by now. Starting in Norway in 1991, regulatory reform gradually spread to Sweden (1996), Finland (1997) and Denmark (2002). In all countries, separation of competitive and monopolistic activities, establishment of independent TSOs and allowing consumers to choose their supplier have been integral parts of reform. With a long tradition of Nordic co-operation, and with development of the jointly-owned power exchange Nord Pool, the Nordic market is now de facto fully integrated, at least at the wholesale level. In this paper, we analyse some issues that are currently high on the agenda in the Nordic countries. Space does not allow a full discussion of the developments of the Nordic market, nor a comprehensive analysis of its functioning. 2 Instead, we limit our attention to a few important issues that have arisen lately, partly as a result of the extreme events of 2002-3. 1 The fifth Nordic country Iceland is not interconnected with the others. The term Scandinavia is inappropriate, as it only encompasses Denmark, Norway and Sweden. 2 See for instance Bergman et al (1999) for a more comprehensive discussion of the earlier developments of the Nordic market. 1

In the second half of 2002, inflow to hydro reservoirs was only 54 per cent of the average of the preceding 20 year period (Bye, Hansen and Aune, 2003). As a result, reservoir fillings were at a record low at the beginning of the lowinflow/high-demand winter season. Foreseeing tighter market conditions, producers began restricting supply in late autumn and prices started to rise. The (daily average) spot price peaked at 850 NOK/MWh in January 2003, two to three times the normal level. High spot prices feed through to consumers, who in some cases faced increases in electricity bills of 50 per cent or more. 3 There was speculation that high prices were the result of abuse of market power, as well as a lack of investment in both generation and transmission in earlier years, and that rationing on a massive scale would be required. As it turned out, no such drastic measures were warranted, as responses from consumers and thermal-power producers balanced the market. Even though prices remained high during most of 2003, market conditions gradually normalised. Some saw the events of 2002-3 as a warning sign, or indeed as outright proof that the electricity market is flawed. Others consider its performance through this period as evidence that the market has reached maturity and is robust enough to withstand even quite extreme shocks. We tend to lean towards the latter view. Nevertheless, the supply shock brought to the surface a number of potential weaknesses that warrant careful analysis and which may eventually lead to further improvements in the regulatory framework as well as in other market institutions. After describing the events of 2002-3 in some detail, we analyse two issues that have attracted considerable attention in the aftermath of these events. The first issue concerns the operation of the retail market; in particular, whether there is sufficient competition on the market, and whether current contractual arrangements are adequate for consumer needs. The second issue concerns generation and transmission capacity; in particular, whether sufficient investment is forthcoming and reasonable levels of supply security and adequacy may be maintained. We end the paper with a discussion of how the implementation of the Kyoto protocol may affect the Nordic electricity market. COPING WITH A SUPPLY SHOCK 4 The development of the electricity market during the winter of 2002-2003 was spectacular, with prices reaching unprecedented levels and a constant threat according to some observers of rationing on a massive scale. We concentrate our attention on events in the Norwegian segment of the market, where effects were at their most extreme. 3 Note that, since many Nordic consumers rely on electricity for most domestic energy needs, incl. heating, electricity bills tend to make up a considerable share of household budgets. For a typical Norwegian household, annual electricity consumption is around 20 MWh (compared to an average of 3.6 MWh in Britain), while the annual bill would amount to around NOK 14,000 (approx. Euro 1,700) at a price of 250 NOK/MWh. 4 This section draws extensively on Bye, von der Fehr, Riis and Sørgard (2003); see also Bye (2003), Bye and Bergh (2003) and Bye, Hansen and Aune (2003). 2

100 90 80 70 60 50 40 30 20 10 0 02.01.02 13.02.02 27.03.02 08.05.02 19.06.02 31.07.02 11.09.02 23.10.02 04.12.02 15.01.03 26.02.03 09.04.03 21.05.03 02.07.03 13.08.03 24.09.03 Per cent 900 800 700 600 500 400 300 200 100 0 NOK/MWh Spot price Hydro stocks Median stocks Figure 1: Spot price and Norwegian hydro stocks, actual 2002-2003 and median 1990-2000 (source: Statistics Norway and Norwegian Water and Energy Authorities) Figure 1 shows the Nord Pool (system) spot price (daily average) and the level of Norwegian hydro stocks 5 from the beginning of 2002 to the end of the summer of 2003. (The average exchange rate for 2002 was 7.5 NOK/Euro.) The figure also shows median hydro stocks over the period 1990-2000. From a low level during the summer of 2002, the spot price rose gradually during the early autumn. This is normal and reflects the fact that limited storage capacity makes it impossible to transfer sufficient water into the high-demand/low-inflow winter season to equate prices over the year. However, towards the end of the autumn the spot price rose steeply and continued to rise well into the winter, when it peaked at around 850 NOK/kWh. The spot price then fell during the late winter and spring, but remained relatively high during most of 2003. The price development reflects the development of hydro stocks. Stocks fell from high levels in the summer of 2002 to record low levels in the following winter. The most obvious reason for this unusual development was the extremely dry hydrological conditions with an almost total stop in inflows to reservoir during the normally wet weeks of the late autumn. 6 5 The Nordic hydro generation capacity is almost entirely located in Norway and Sweden. The Swedish hydro stocks followed a parallel development to the Norwegian. 6 Inflow to reservoirs is at its minimum in the winter season when most precipitation is in the form of snow, and it reaches its maximum in the late spring and early summer when the snow melts. Autumn is usually rather wet, with almost all precipitation in the form of rain, and hence inflow is relatively high in this period also. 3

9000 8000 7000 6000 5000 4000 3000 2000 1000 0 1 5 9 13 17 21 25 29 33 37 41 45 49 We e k Year 2002 Yearly average, 1970-99 Figure 2: Weekly inflow (GWh) to Norwegian hydro reservoirs (Source: Norwegian Water and Energy Authority) As shown in Figure 2 which compares weekly inflow to Norwegian hydro reservoirs during 2002 with yearly averages over the period 1970-99 the year 2002 actually started out as rather wet. Until the summer, inflow was consistently above the historical average; indeed, in the 24 first weeks of 2002 inflow was 14 TWh, or nearly 20 per cent, above average. However, in early autumn inflow fell below normal levels and from October onwards it more or less dried up completely; during Weeks 38-48 inflow was 9,3 TWh below average. It has been argued that the fall in hydro stocks could have been avoided if generators had restricted supply at an earlier stage. However, with the very high levels of stocks in the early autumn there was apparently a real risk that with a wet autumn reservoirs would have become so full that water would have been spilled and wasted. 4

100 90 80 70 Per cent 60 50 40 30 20 10 0 1 5 9 13 17 21 25 29 33 37 41 45 49 Week Act ual Project ed min Project ed max Project ed mean Figure 3: Norwegian hydro stocks 2002, actual and projected from Week 30 (source: Statistics Norway and Norwegian Water and Energy Authorities) Figure 3 shows actual total hydro stocks, as well as projected levels as seen from Week 30. With maximum inflow reservoirs would have become completely full. Given differing levels of stocks, inflow patterns and constraints on output, the risk that any individual reservoir would be filled up was likely to be higher than suggested by these figures. As it turned out, such a wet outcome did not occur; instead, inflow and hydro stocks were in fact close to the lowest projected level. This evidence would seem to be consistent with a view that generators acted rationally upon information available at the time. On the other hand, the evidence is also consistent with the view that hydro generators were overly anxious to tap their reservoirs, possibly with the intention of pushing up prices, which, at the time, looked to remain at modest levels during the winter. In any case, the mere suspicion that such anti-competitive practices were pursued may well have been influential in determining the tough stance that the Norwegian government took on attempts by the dominant (and state-owned) generator Statkraft to buy up more of its smaller Norwegian rivals. Another possible explanation for the price spike popular among a number of commentators was that a lack of investment over a long period of time had eventually lead to under-capacity and a consequent imbalance between supply and demand. It is certainly true that investment levels had been low for a number of years, but whether this was a sign of market imperfection is questionable. We discuss the investment issue in some detail below. Here we only point to the fact that even though prices reached record levels during 2002-2003, they had been well below levels that would make new investment profitable for most of the preceding 10-15 years. Forward contract prices were also low. For instance, in 1999 prices in forward contracts covering the year 2002 were around 150 NOK/MWh, well below the 200 NOK/MWh estimate at the time of unit costs for new gas-fired plants. Forward prices gradually 5

increased in subsequent years, although in 2002 prices covering the year 2006 were still not higher than 180 NOK/MWh. There consequently seems to be little support for the view that generators had not seized on profitable investment opportunities. A criticism along similar lines concerned investment in transmission capacity. 7 Some commentators argued that lack of investment in transmission capacity both within and between the Nordic countries had allowed the development of areas where severe shortages were bound to arise. Again, it is true that there had been relatively little investment in transmission capacity over a number of years, and that the 2002 supply shock lead to long periods of time in which the market was effectively segmented. In particular, import-capacity to Norway, and export capacities from the thermal-dominated systems in Denmark and Finland, were constrained during much of the winter of 2002-2003. Nevertheless, it is not clear that this was a sign of deficiencies in the transmission network. On the one hand, the Nordic countries are highly interconnected, and most of the time the market is fully integrated at a single market-wide price. On the other hand, bottlenecks are becoming more frequent, and, in combination with increasing levels of market concentration, there is a worry that the result may be imperfect competition and inefficient market outcomes. We return to this latter issue below. We also return to the issue of transmission investment in connection with our discussion of supply security and adequacy. 7 A completely opposite, but nevertheless quite popular view was that high prices in Norway were caused by excessive interconnection with neighbouring countries, leading to a drain on hydro resources and consequent shortage prices. While there was certainly some truth in this view, it seemed to overlook the fact that in the face of a severe, negative supply shock prices would reach even higher levels without access to imports. We do not discuss the possible (protectionist) implications of this view. 6

700 600 500 NOK/MWh 400 300 200 100 0 2004:1 2003:4 2003:3 2003:2 2003:1 2002:4 2002:3 2002:2 2002:1 2001:4 2001:3 2001:2 2001:1 2000:4 2000:3 2000:2 2000:1 1999:4 1999:3 1999:2 1999:1 Households Manufacturing industries Service industries Figure 4: End-user prices (excl. taxes and network tariffs), quarterly observations, 1999-2004 (source: Statistics Norway) The high spot prices feed through to end-user prices. Figure 4 shows quarterly observations of the energy element in average retail prices from early 1999 to early 2004 for households, services industries and manufacturing industries, respectively. Retail prices shot up at the end of 2002 and soon reached unprecedented levels. Prices paid by households tended to increase more than those paid by industrial consumers. The difference seems to be explained by the different composition of contracts in the various segments of the market. Most household consumers have so-called variable-price contracts, according to which retailers can change the price with a few weeks notice. As of the first quarter of 2003, 85 per cent of household consumers had such contracts; another 7 per cent were on spot-price contracts (with the retail price directly linked to the Nord Pool spot price) and only 8 per cent on fixedprice contracts (see Figure 5). This is different from industrial consumers, especially in the manufacturing industries, who tend to rely more on long-term, fixed-price contracts. In the first quarter of 2003, 55 per cent of consumers in the manufacturing industries and 22 per cent of consumers in service industries had fixed-price contracts. The corresponding figures for spot-price contracts were 35 and 53 per cent, and for variable-price contracts 10 and 24 per cent. Consequently, industrial consumers were less exposed to price increases than were households. 7

100 % 90 % 80 % 70 % 60 % 50 % 40 % 30 % 20 % 10 % 0 % 2002:1 2002:2 2002:3 2002:4 2003:1 2003:2 2003:3 2003:4 2004:1 2002:1 2002:2 2002:3 2002:4 2003:1 2003:2 2003:3 2003:4 2004:1 2002:1 2002:2 2002:3 2002:4 2003:1 2003:2 2003:3 2003:4 2004:1 Households Service industries Manufacturing industries Variable price Spot price Fixed price Figure 5: Contract shares, quarterly observations (source: Statistics Norway) There was a general move from variable-price contracts to fixed-price contracts in the wake of the events of the winter of 2003. This trend seems now to be reversed (maybe because the memory of the price spike is starting to fade?). Interestingly, there is little interest among household consumers as opposed to industrial consumers for so-called spot-price contracts, in which the retail price is linked to (an average of) the Nord Pool Elspot price. There is ample evidence that spot-price contracts perform consistently better than variable-price contracts in the longer term; in particular, it would seem that competition between suppliers of variable-price contracts is not always entirely effective. The reason households have not embraced spot-price contracts may be (an erroneous) view that these contracts, being linked to the highly variable spot price, are somehow more risky than variable-price contracts. Increases in end-user prices had a considerable impact on demand. Roughly speaking, demand may be seen as consisting of three segments: the very flexible boiler segment (approx. 5% of the total), the heavily-contracted power-intensive industry (approx. 30%) and the rest (approx. 65%). Demand from the boiler segment which can easily switch between oil and electricity fell sharply when prices started to rise in October 2002 and remained low during the winter; all in all, electricity consumption by boilers over the period November 2002 to May 2003 was around one third of that of the corresponding period in 2001-2002. In the energy-intensive industries, some plants stopped production, but the overall response was relatively small, probably less than 5 per cent (Bye, von der Fehr, Riis and Sørgard, 2003). 8 In the remaining segment households and other industry temperature-adjusted demand fell by 7 per cent over the 8 The response seems small in comparison to estimated reservation prices for the power-intensive industry. The industry has the option of moth-balling plant and selling power, obtained of preferential terms determined by the Norwegian Parliament, in the spot market (Bye and Larsson, 2003). The lack of response may be due to uncertainty about price developments, long stop and start-up times, and the risk of undermining popular and political support for the industry. 8

November-May period compared to the year before; given an average increase in end-user prices of 30 per cent, this corresponds to a price elasticity of 0.23. The experience in Norway may be contrasted with that of the other Nordic countries. Although wholesale prices moved more or less in parallel, retail prices were much less affected in these countries. This would seem to be explained by the fact that retail markets differ, particularly in the availability and composition of contracts, but also in market structure and the extent of competition. In Denmark and Finland, where fixed-price contracts dominate, domestic consumers were much less exposed to price increases than in Norway. In Sweden, there is a greater variety of contract types, although the incidence of long-term, fixed-price contracts is higher than in Norway. Moreover, as we discuss below, there seems to be less competition among Swedish than among Norwegian retailers. Also, in Sweden retail prices reacted much less than in Norway. As a result, the demand response was much less in these countries than in Norway. MARKET INTEGRATION AND RETAIL COMPETITION The events of 2002-2003 cast light on a number of potential problems, including concentration and scope for market power, contractual coverage and exposure of consumers to price risks, investment and its impact on supply security, as well as bottlenecks in the transmission network and segmentation of the market. Below we focus on the retail market (this section) and issues concerning supply security and adequacy (next section). We also discuss how the introduction of new, market-based instruments for regulation of environmental emissions may impact on the performance of the electricity industry (penultimate section). The establishment of Nord Pool and the elimination of border tariffs between the Nordic countries were key elements in a strategy aiming at an integrated Nordic market for electricity. The success of this strategy may be measured by the degree of wholesale and retail price equalisation between the different price areas. 9 Obviously, an uneconomically large transmission capacity would be required if transmission constraints were to be eliminated, enabling wholesale prices to be equalised across all areas at all times. However, significant and persistent deviations between area prices would imply that the Nordic market consists, in effect, of a set of national or regional electricity markets. As we show immediately below, the wholesale market appears to be strongly integrated, with prices in different areas diverging for shorter periods only. However, as mentioned above, retail market prices reacted very differently 9 Whenever interconnector capacity constrains power flows, the Nord Pool market is divided into two or more price areas. Sweden is always treated as a single price area, and the same applies to Finland. This is because congestion in the national transmission systems is managed by means of socalled counter-trade in these countries. In Denmark, the eastern and western parts of the country are physically separated and hence there are always two price areas East and West. In Norway, segmentation of the market is part of the handling of transmission constraints and the country may be divided into two to five price areas, depending on the demand-supply configuration. 9

across the Nordic countries to the 2002-2003 increase in wholesale prices: while they shot up in Norway, the reaction was much more subdued in Sweden, and in Denmark and Finland retail prices hardly changed at all. There are also considerable differences in the level of retail prices, even when one corrects for differences in taxes and network tariffs. Some of these differences can be explained by differences in regulatory regimes. We concentrate our attention on Norway and Sweden, where regulations are similar, but where retail markets nevertheless seem to perform quite differently. WHOLESALE PRICES Table 1 displays the annual averages of the Elspot system 10 and area prices over the period 1996-2003. The figures indicate that deviations between system and area prices have been quite small. Except for the years 2000 and 2003 when the supply of hydropower, especially in Norway, significantly deviated from normal levels the Nordic electricity market appears to be reasonably close to being a single market. Table 1: Elspot system and area prices 1996-2003, annual averages, NOK/MWh (source: Nord Pool) 1996 1997 1998 1999 2000 2001 2002 2003 System 253.6 135.0 116.4 112.1 103.4 186.5 201.0 297.5 Norway, Oslo 256.7 137.5 115.7 109.2 97.7 186.0 198.5 301.7 Norway, 251.2 133.0 116.2 119.5 100.7 188.6 200.2 295.7 Tromsø Sweden 250.6 135.0 114.3 113.1 115.5 184.2 206.3 292.8 Finland - - 116.3 113.7 120.7 184.0 203.8 277.9 Denmark, - - - 113.7 120.6 191.2 190.7 268.3 West Denmark, East - - - - - 189.7 213.7 291.7 However, small discrepancies between annual averages may hide shortterm variations in different directions and is thus only a very crude indicator of the degree of price equalisation. As can be seen from Table 2, the number of hours in which the entire market has been integrated that is, when the system price has been exactly equal to all area prices is below 60 per cent. The corresponding figures for Finland-Sweden are in the range 75-100 per cent, for Norway (Oslo)-Sweden 70-85 per cent (except in 2000) and for Finland- Norway (Oslo)-Sweden 60-85 per cent (except in 2000). Moreover, the share of the time when Sweden has been a price island is in the range 0-5 per cent. Thus, in terms of wholesale price equalisation the Nordic electricity market would seem to be reasonably well integrated. 10 The system price is calculated under the assumption that there are no transmission constraints. Actual trade is carried out at the system price only when transmission constraints are not binding. 10

Table 2: Number of hours with complete Elspot price equalisation 1997-2002 (source: www.seef.nu) 11 1997 1998 1999 2000 2001 2002 Number of hours 5 201 3 825 3 788 1 703 4 487 3 076 Share of time, % 59.4 43.7 43.2 19.4 51.2 35.1 RETAIL COMPETITION There is complete market opening (i.e. full retail competition) in all the Nordic countries. In some of the countries, such as Sweden, a household consumer may even buy electricity from suppliers in any Nordic country. Given this, the pre-tax retail prices should not differ very much between the four Nordic countries. However, there are obstacles to transactions between suppliers in one country and households and other small customers in other countries. For instance, in order to be able to supply electricity to a Swedish customer located in Stockholm a non-swedish supplier needs to buy electricity in the Stockholm price area. Moreover, as a buyer in the Stockholm price area the supplier needs to have a contract with a so-called balance responsible party (as well as in the home country). As a result of such obstacles, only a sub-set of all retailers in the Nordic countries is actually competing on the Swedish market. In addition, only a subset of all Swedish retailers competes outside the geographical area in which they are located. Corresponding situations prevail in the other Nordic countries. Consequently, retail electricity prices need not necessarily be equalised across national borders. In order to shed some light on this issue, we have compared retail prices in Norway and Sweden, the two countries that have operated a common wholesale market since 1996. Retail prices differ between households for several reasons. One reason is related to non-linearity of price schedules and annual consumption patterns of households. Thus prices paid by households living in a single-family houses with electric heating typically consuming around 20 MWh per year is lower than the prices paid by households using electricity only for lighting and electrical appliances, consuming as little as 2 MWh per year. 12 Another reason for retail price differences is that customers can choose between fixed-price contracts (with different duration) and variable-price contracts, and that the prices charged for these contracts may deviate in the course of year. In both Norway and Sweden, the default contract the type of contract that applies for customers who have not actively chosen to change supplier or signed a new 11 Presentation by Dr. Niklas Strand, Swedish Competition Authority, at the Swedish Association for Energy Economics. 12 To some extent this pattern is surprising. On the one hand, it is likely that economies of scale may motivate some quantity discounts. On the other hand, it is obvious that households with electric heating consume electricity primarily during the winter period when peak generation capacity is used and spot prices are high. As spot prices are sometimes very high during the winter season, one would expect that the cost of offering the customer insurance against winter price spikes, which is what the retailer does, would be rather high. 11

contract with the old supplier is a variable-price contract; that is, a contract that allows the supplier to adjust the price (after notifying the customer) and so, in effect, pass on cost increases to customers. A third reason for retail price differences is that individual retailers may adopt different market strategies and offer different combinations of prices and services. In Table 3 and Table 4, annual averages household prices over the period 1997-2003 are displayed for, respectively, Norway and Sweden. The numbers reflect averages of prices offered by all retailers to households with an annual consumption equal to 20 MWh. Table 3: Retail prices, net of taxes, for 20 MWh household consumer according to contract type in Norway 1997-2003, NOK/MWh (Source: Statistics Norway) 1997 1998 1999 2000 2001 2002 2003 Normal n.a. 160 151 141 210 205 454 Spot n.a. n.a. 131 123 193 193 323 1 year n.a. n.a. 152 144 189 195 287 fixed Average 210 162 152 141 206 203 414 Table 4: Retail prices, net of taxes, for 20 MWh household consumer according to contracts in Sweden 1996-2003, NOK/MWh (Source: Statistics Sweden) 1997 1998 1999 2000 2001 2002 2003 Normal 259 251 244 218 225 296 447 1 year fixed 178 181 256 397 2 year fixed 177 184 253 351 3 year fixed 182 186 252 324 It is immediately clear that Norwegian and Swedish retail prices differ significantly. No law of one price is visible, and the retail prices have been significantly higher in Sweden during most of this period. Thus, with regard to the household market, there seems to be two national rather than one integrated Norwegian-Swedish electricity market. Annual variations of the Norwegian retail prices are quite significant, but also relatively well correlated with variations in Elspot prices (comp. Table 1 and Table 3). Thus retail prices fell during the wet period 1997-2000, increased in 2001 when precipitation was normal, and skyrocketed in 2003 when Nord Pool prices were extremely high. In the case of Sweden, however, the correlation between Elspot and retail prices was rather weak during the period 1996-1999. Instead, average retail prices remained high in 1998 and fell only marginally in 1999, while a quite significant reduction took place in 2000 and 2001. 12

The most obvious explanation for these differences between the development of retail prices in Norway and Sweden are related to switching costs at the household level. In Norway, a system of profiling was adopted from the outset and customers could switch to another supplier at no cost; more specifically, there were no requirement to install interval or real-time meters, nor any charges levied. The option of changing supplier at no charge has led to a sizeable swing of customers away from the old local suppliers to new suppliers that offer electricity at lower prices. In other words, there were no significant switching costs protecting the old suppliers from competition. In Sweden, on the other hand, costly real-time metering and reporting were required for consumers wanting to change supplier. These regulations were in effect until November 1999, and as a result few households changed supplier or renegotiated their contract with the old supplier. After November 1999, a system of profiling has been in place. Consequently, there was a significant reduction of switching costs in Sweden at the end of 1999. The figures in Table 4 suggest that retail competition was rather modest as long as switching costs were high. At the same time, the figures indicate that the reduction of switching costs opened up a significant competitive pressure on retail prices. Nevertheless, even with a regulatory regime more conducive to competition, retail prices continue to be higher in Sweden than in Norway, in spite of the fact that retailers procure electricity at the same well-integrated wholesale market. A natural hypothesis is that this is a result of market power. MARKET STRUCTURE Traditionally, the major generating companies in Sweden have had rather small shares of the retail market. Thus, although Vattenfall for a very long time has been the single biggest retailing company, its share of the retail market was only around 15 per cent in the middle of the 1990 s. However, in the last few years the major generating companies Vattenfall, Sydkraft and Fortum (formerly Birka) have bought majority or minority shares in a number of small and medium-sized retailing companies. In most cases, sellers have been towns and municipalities. Moreover, some of the independent retailing companies that entered the market in 1999, such as the Norwegian oil and gas company Statoil, have since left the market. As a result of these developments the number of retailing companies has been reduced, and the big three have become dominant players on the retail market. For instance, if we include retailing companies in which generators own minority shares, Vattenfall is currently serving around 30 per cent of all Swedish customers, while the corresponding number for the big three is around 70 per cent. Similar numbers apply to shares of volumes of electricity delivered to final consumers. In Norway, power generation has traditionally been much less concentrated than in Sweden. Except for the state-owned company Statkraft, accounting for some 30 per cent of total Norwegian power generation, generators are small, with market shares of 5 to 6 per cent or lower. As Statkraft and the second largest company, Norsk Hydro, almost exclusively serve industry and other 13

businesses on long-term contracts, the retail and household market in Norway has been much less concentrated than in Sweden. In recent years, however, significant changes in the Norwegian power sector have taken place. In particular, many companies in local-government ownership have been turned into limited-liability companies, often as part of a process leading up to the sale of ownership interests. Also, larger regional power companies have been established, partly by acquisition and partly through mergers. Furthermore, Statkraft has acquired stakes in several Norwegian power companies. Foreign companies have also acquired some ownership interests in Norwegian companies (notably in grid management and operations and in power retailing). In spite of these developments, however, the Norwegian market, both at the wholesale and retail level, remains less concentrated than its Swedish counterpart. MARKET POWER AND PRICE DISCRIMINATION So far, the retailing segment of the Swedish electricity supply industry has not been very profitable, and several entrants to the market have had to leave after having suffered significant losses. On the whole, it seems that the costs of the retailing business have been severely underestimated; in particular, it seems that the exposure to price- and quantity risks have been more costly than expected. 13 But in 2002 the established retailers were able to implement an increase in trade margins without attracting new entrants to the market. The question then is why the big three have grown while several independent retailers have left the market. A possible explanation may be that integrated generation-retailing companies are more efficient than independent retailers. In Sweden, legal separation between retailing and distribution is required. This provision has paved the way for significant integration between generation and retailing. It seems that the integrated companies have a competitive advantage in relation to independent retailers stemming from the lack of efficient markets for hedging against area price and quantity risks. Thus, while all retailers may suffer from a combination of unexpectedly high area prices and consumption levels, the extra costs in the retailing business become extra revenues in the generation business for the integrated generation-retailing companies. If these hypotheses could be verified they clearly point at an unexpected effect of the legal separation requirement in Sweden. The intention was to stimulate retail competition by preventing cross-subsidisation between distribution and retailing. This objective may have been attained, but recent developments suggest that allowing vertical integration between generation and retailing also made it possible to exercise market power in the retail market. 13 Retailers can hedge Elspot system price risks at Eltermin, and the liquidity of these instruments (futures, forwards and options) traded at Eltermin is high. However, retailers in Sweden have to buy electricity at the relevant Swedish area price, and opportunities for hedging idiosyncratic area price risks are not very well developed. Moreover, retailers can only hedge price risks for a fixed number of MW:s per hour, while their customers do not have to commit to a certain quantity. Thus retailers are exposed to the risk of having to buy extra electricity at spot-market price during hours when their customers have an unexpectedly high level of consumption. 14

There also seems to be an element of price discrimination in Swedish retail prices. As seen in Table 4, there is a considerable spread between prices in the so-called normal contract and prices in fixed-prices contracts, or between prices paid by customers with default contracts and customers who have actively switched to a new contract; in particular, prices in default contracts are higher than those in fixed-price contracts. This suggests that Swedish retailers are able to price-discriminate against the default-contract customers. After all, by refraining from actively choosing another type of contract these customers have demonstrated that they are not very price-sensitive and it must be tempting for suppliers to utilise this information. SUPPLY SECURITY AND ADEQUACY A notable consequence of regulatory reform in the Nordic countries and, indeed, one of its most important successes has been the almost complete halt to investment in generation and, to a lesser extent, in transmission and distribution. GW 100 90 80 70 60 50 40 30 450 400 350 300 250 200 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 GWh 2002 Installed capacity Gross consumption Maximum system load Figure 6: Capacity, consumption and system load (source: Nordel) Figure 6 shows installed generation capacity in the Nordic market since 1980. Over the ten year period preceding the first regulatory initiative, from 1980 to 1990 (the year before the new Energy Act took effect in Norway), generation capacity grew by 30 per cent. Over the next ten year period, from 1990 to 2000, installed capacity grew by a mere 3 per cent. Indeed, in 2003 installed capacity was more or less the same as in 1996 the year when 15

regulatory reform was introduced in Sweden: a fall during 1998-9 was only reversed by subsequent increases in recent years. 14 The stagnation in capacity growth cannot be explained by development of demand. Admittedly, demand did not grow at the pace experienced in the early 1980s and before, but gross consumption continued to grow at a more or less constant rate of 1-1.5 per cent per year (see Figure 6). As a consequence, overcapacity inherited from the pre-reform era has gradually being reduced, if not entirely eliminated. Comparing generation capacity and maximum system load, we find a similar picture, although the trend is perhaps not as pronounced. Nevertheless, the capacity margin defined as the excess of installed generation capacity over maximum load reached it lowest level in 2001. The development of generation capacity must be seen in relation to stricter regulatory policies, arising mostly from environmental concerns. Although a considerable amount of undeveloped hydro capacity still remains, it is unlikely that many more new hydro sites will be developed. Nuclear power traditionally important in both Finland and Sweden has long been viewed with great scepticism (although the Finnish Parliament recently approved the building of a new nuclear power plant). Increasing concerns about air pollution has lead to strict regulations, not only on coal- and oil-fired power plants, but also gas-fired plants. Notwithstanding the importance of environmental regulation, it would seem that regulatory reform with the abolition of monopoly rights, integration of markets and development of market places has been the most important factor in determining generator investment behaviour. Market-based competition not only reduced prices, but also turned the focus of market participants towards profitability. Indeed, in an industry traditionally committed to a public service ethos, regulatory reform legitimised a more capitalist attitude. The greater emphasis on profits lead to company restructuring and mergers, as well as to increased efficiency (eg., employment in the Norwegian electricity industry fell from almost 20,000 in 1993 to below 13,000 in 2002). Specifically, electricity supply companies increasingly required returns on investment in line with those obtained in other industries. 14 This movement seems to be explained largely by plants that were first moth-balled and then reopened. 16

18 % 16 % 14 % 12 % 10 % 8 % 6 % 4 % 2 % 0 % 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Electricity industry Manufacturing industries Figure 7: Return on capital, 1990-2002 (source: Statistics Norway) Figure 7 shows return on capital in the Norwegian electricity industry since 1990, the year before deregulation. 15 Over this period, rate of return has averaged 5.5 per cent, only half that achieved in the Norwegian manufacturing industry. Due to the existence of a resource rent in a hydro-dominated electricity industry, we would expect the average rate of return to exceed that on a marginal plant. Consequently, it seems relatively safe to conclude that new investment in generation capacity could not have achieved reasonable levels of profitability over this period. Nevertheless, even though low levels of investment seem to have been a rational response to prevailing market conditions, the question remains whether investment will be forthcoming as the earlier over-capacity is eroded and market conditions become tighter. In other words; are there reasons to believe that market imperfections will inhibit investment and undermine the future performance of the industry? In order to answer this question, it is important to distinguish between two related, but nevertheless entirely different, concepts: supply security or balancing consumption and generation on a continuous basis within existing capacity limits; and supply adequacy or ensuring optimal capacity investment by balancing willingness to pay for new capacity against its cost. In other words, while the supply-security issue is short term and mainly concerns system operation, the supply-adequacy issue is long term and concerns the evolution of capacity in relation to consumption. In interpreting the balance of consumption and generation one must be aware of transmission constraints 15 Rate of return is measured according to National Accounts as operating surplus relative to the value of capital employed. 17

that limit the range of generators that can meet demand in a given location. Below, we discuss these issues in some detail. SUPPLY SECURITY: BALANCING DEMAND AND SUPPLY As is well known, the problem of continuously balancing consumption and available generation arises from specific features of electricity markets, including the need for electrical equilibrium at all times, unexpected variations in demand and supply, limited possibilities for establishing and transmitting adequate price signals to market participants on a continuous basis, and limited short-run response by market participants to price signals. The gain from increasing supply security is associated with a reduction in the costs of rationing. A rationing event occurs when, at prevailing prices, the desired demand and/or supply of market participants cannot be satisfied and hence their decisions have to be constrained. For example, if demand exceeds supply at prevailing prices, either additional supply (if available) has to be ordered onto the system, or the consumption of one or more consumers has to be forced down. Rationing may be voluntary or involuntary. One example of a voluntary rationing agreement is contracts for operational reserves, by which the system operator obtains the right to call on additional supplies when needed. Another example is long-term load-shedding contracts between consumers and their suppliers (or between consumers and the system operator), by which consumers, upon conditions that have been agreed in the contracts, can be called on to reduce their load. Involuntary rationing typically occurs by zonal interruptions of power supplies. Supply security cannot be ensured by capacity investment alone, although a larger capacity may reduce supply security problems. Optimal utilisation of existing generation capacity whatever its level involves setting prices that allow for the highest possible degree of capacity utilisation while at the same time securing sufficient reserve margins. Having more capacity available essentially means that prices will remain at lower levels so that more demand is encouraged and a high level of capacity utilisation is ensured. Conversely, when capacity is limited prices will increase so as to reduce demand. Provided generation capacity is optimally used, on longer time scales demand for capacity will follow available capacity and system reserves will not be directly linked to total capacity. Obviously, in the Nordic system, with a large incidence of hydro power, there will be sustained periods of time in which the energy balance is tight. In such situations, a continued balance between demand and supply requires that prices rise, as happened during the winter of 2002-2003. Note that the price rise would have been less in that event if more of demand had been exposed to the actual cost of electricity: the reliance on fixed-price contracts in large segments of the market exacerbated the supply deficit and pushed wholesale prices higher than they otherwise would have been. On the other hand, it is conceivable that one may in the future experience an even more severe reduction in inflows, with even higher prices as a result. Nevertheless, if, by adjusting price levels, demand can be scaled to be (on average) aligned with existing capacity, the supply security problem is not associated with the level of demand (relative to existing capacity) per se but rather with variations in demand (and capacity availability). 18

Given this, it must be noted that, in a hydro system, even if the energy balance is tight there is generally a large amount of (power) capacity available. Since hydro plants are typically constructed so as to be able to handle peak inflow levels, and adjustment of output is extremely cheap, the ability to deal with short-term variations in the system is not necessarily impaired in dry periods. To sum up, it is not at all clear that the Nordic market is particularly vulnerable to supply insecurity, even if market conditions were to become tighter in the future. Indeed, the technology mix and the mere size of the market would seem to allow for a very high degree of supply security. Moreover, regulations are in place which should provide Nordic system operators with the tools they need to balance the system. Take the Norwegian TSO Statnett as an example. Firstly, rights and responsibilities have been clearly set out in specific regulations; in particular, Statnett is responsible for system balance and has the right to make the necessary contractual arrangements with market participants to achieve this task. Secondly, market-based institutions have been set up to achieve balancing in a cost-effective manner. Most importantly, Statnett runs a (near-to) real-time balancing markets in which balancing services are sourced on a short-term basis. Furthermore, at regular intervals Statnett procure strategic reserves, partly in the form of contracts for interruptible demand. All in all, these instruments should be sufficient to guarantee that balancing is achieved at reasonable costs. Studies have indicated that the efficiency of system operations would be further enhanced by tighter co-operation between of system operators (Bjørndal and Jørnsten, 2001; see also von der Fehr, Hagen and Hope, 2002). Developments in this direction have recently taken place, with integration of the national balancing markets (Nordel, 2003). However, these efforts are probably not sufficient, and further gains may be had from optimising the system as a whole, including lower overall reserve margins and increased transmission capacity. SUPPLY ADEQUACY: OPTIMISING CAPACITY INVESTMENT The supply adequacy problem essentially consists of two elements, namely ensuring an optimal level of overall generation capacity and an optimal mix of different generation technologies. An optimal level of overall capacity is characterised by equality between willingness to pay for new capacity and the cost of such capacity. In other words, a situation of under-investment in generation capacity would be characterised by investment not forthcoming even though the willingness to pay for the associated increase in output is more than sufficient to cover its cost. Similarly, a situation of over-investment would be characterised by the cost of marginal capacity units exceeding consumer willingness to pay for the associated output a situation well known from the history of the Nordic electricity industry. An optimal mix of generation technologies is characterised by the minimisation of costs of satisfying a given consumption profile. Cost-efficient 19

operation requires a mix of technologies with different variable to fixed cost ratios. At one extreme, low-variable/high-fixed costs technologies such as hydro, nuclear and conventional thermal operate continuously as base-load units; at the other extreme, high-variable/low-fixed costs technologies such as small gas- or oil-fired units are used for demand peaks only. An optimal capacity mix balances the gains from reducing variable operating costs by having more base-load units available against the higher fixed costs of such units. At the moment, there would seem to be two major concerns regarding supply adequacy in the Nordic market: firstly, whether investment incentives are sufficient to allow overall capacity to expand at a reasonable pace, and secondly, how the generation park will be affected by the introduction of new environmental regulation initiatives. Here we focus on the first issue of general investment incentives and leave the latter issue for the next section. In the hydro-dominated Nordic market, wholesale prices swing from year to year, depending upon hydrological conditions. However, these fluctuations in prices tend to average out. Indeed, judging from prices in forward contracts which are traded up to three years ahead on Nord Pool expected prices tend to be quite stable, evolving slowly in reaction to changes in underlying fundamentals. As at late 2004, contracts for two and three years ahead are trading at around 250 NOK/MWh (31 Euros/MWh), a level that would make investment in conventional gas-fired plant approximately break even. It would seem that these prices are in fact sufficient to attract investor interest. The Finnish government and the EU Commission recently approved a new 1.6 GW nuclear power plant, expected to be in operation from 2009. In Norway, Statoil recently unveiled plans to build a 860 MW gas-fired plant and a 280 MW co-generation unit at its industrial plants at Tjeldbergodden and Mongstad (landing points for North-Sea gas). Gas-fired generation has been a source of controversy in Norway leading to the downfall of at least one government and it remains to be seen whether the Statoil projects will be approved. However, whatever the outcome, it would seem that generation investment in the Nordic market is not so much a question of commercial, as of political, will. 16 Transmission investment has been fairly limited for a number of years. There seems to be two main issues: first, how regulation of the individual TSO affects investment incentives and, second, the ability of national TSOs to solve co-ordination problems associated with investment in interconnector capacity. In the case of the Norwegian TSO, Statnett, there appears to be no lack of will to invest. Given that costs of new investment will in effect be passed on to network users, Statnett has shown considerable willingness to undertake new projects. Were it not for the more restrictive view taken by Norwegian regulatory authorities, more transmission capacity would indeed have been 16 Uncertainty surrounding the future of the Swedish nuclear capacity is another important example of how the political process interacts with investment incentives. Investment in renewables, such as wind power and generation based on burning of biomass, is critically dependent upon government support. 20